Z decays to b-jet pairs: submission to NIM approved – part 1 January 11, 2008Posted by dorigo in news, personal, physics, science.
And after all the paper seminar went smoothly. A 20 minutes talk by Julien at the CDF weekly meeting completed a very long process within our collaboration, and a green light was set on for a paper to be sent to the editors of Nuclear Instruments and Methods in Physics Research for prompt publication.
Julien’s slides are a bit technical, but I will nonetheless paste a few of them on this blog tomorrow, just for the record. Before I do, however, let me tell you a little anecdote on how the whole thing started in a separate post, id est, here.
Exactly twelve years ago I was a freshman graduate student starting to look for a suitable topic for my PhD thesis. I was interested in the four-jet dataset that CDF had just finished collecting in Run I: 110 inverse picobarns of collisions had been filtered by a trigger designed for the top pair production search in the all-hadronic final state, the one arising when each top quark produces three energetic partons, and you obtain six jets from the decay, as shown in the graph on the right.
About a million events with four or more energetic jets were in my hands. The data were being exploited by my group in Padova, in fact, for the task it had been designed for: one year later I would become one of the authors of the http://www-cdf.fnal.gov/physics/preprints/cdf4025_top_allhad_prl.ps.gzall-hadronic top observation paper. But those data -for the most part due to our dreaded background, the production of multijets by strong interactions (QCD for parvenues, an acronym of Quantum Chromo-Dynamics)- made a fascinating sample regardless of its top content: QCD was still not understood at the level it is now, so comparing the kinematics of those events with the predictions of a simulation was by itself a quite informative exercise.
Far more interesting, however, was to select the few events where at least two of the jets contained a b-tag (the signal that the jet had originated from b-quark fragmentation): I remember we had only about a thousand of those, and I started to play with them to understand what kind of physics could be extracted from that very special subset.
At the time, the search for a Higgs boson in the associated production mode, , was considered feasible in four-jet events, despite the huge QCD multijet production cross section. The trick was to understand how b-quark pairs were produced by background processes: gluon splitting and direct production mechanisms. Stuff for another post. Also searches for new physics, like supersymmetry, were quite promising in a fully hadronic final state. Many papers would appear in the forthcoming years on those topics, in fact.
To me, computing invariant masses of pairs of b-quark jets, jet triplet masses, combined momenta, and similar things was an exciting occupation. I knew the data could contain Higgs bosons, supersymmetric particle decays, diboson production: it was like a unexplored mine. I would think of a new observable quantity in the morning on my way to work, code it in fortran in the morning, obtain a plot comparing data to simulation before lunch, and think about it in the afternoon.
It was the dijet mass of pairs of b-quark jets the one which triggered my interest the most: if a certain pseudo-motivated set of cuts were applied ot the data, the bb-mass distribution showed a nagging bump at about 90 GeV, and I could not figure out whether it was a statistical fluke or a hint of the presence of a few decays – a signal that should of course be there, although very small compared to the QCD background.
Z decays to jets at a hadron collider ? Surely I must be joking, was the general consensus among my colleagues in Padova. They were wrong in general, but they were also right in the particular case at hand: a signal cannot be extracted from a four-jet dataset at a hadron collider, no matter how hard you try. The problem is, Z bosons are by themselves quite hard to extract from two-jet events, but when you move to a dataset of events containing two additional jets you end up favoring backgrounds rather than the signal you are after!
Indeed, a study of signal characteristics convinced me that the two b-jets from the boson decay are not usually accompanied by additional jets. The latter are the result of the hadronization of energetic quarks or gluons radiated from the partons producing the collision, or those coming out of it. In events both initial () and final state () are quarks, which carry a smaller color charge than gluons, and emit QCD radiation less frequently. On the contrary in the background processes, when a gluon is the originator of the pair of b-quarks mimicking the Z decay signature, the initial state is most likely gluon-rich, resulting in a higher chance of additional jets being radiated. The end result ? If you are going to extract the small signal from hadronic collider data, you have better veto additional jets, not enforce their presence!
Despite my sitting on the wrong sample, the spark had been thrown to the haystack. I had convinced myself that it would have been quite interesting to extract the Z boson from dijet events, in some other suitable sample of data collected by CDF. This was a challenging task, for a start: nobody had done it before, as only a combined signal of the hadronic decay of W and Z bosons had been detected by UA2 back in 1987 with a dedicated run and a trigger designed exactly for that purpose. Those were the years when W and Z bosons had not been studied in detail yet, and seeing their hadronic decay carried a considerable scientific value.
In 1996 nobody had demonstrated yet that dijet resonances could be exploited for new particle searches, such as those forthcoming for the Higgs boson decay. So it must be of high scientific value, I reasoned, to demonstrate that at least the Z could be seen to decay to hadrons: the production rate of the Z is at least a thousand times larger than that of the Higgs, after all!
There were additional bonuses. A measurement of the decay would check the efficiency of tagging b-quark jets, by comparing predicted and measured rates; it would allow to check the scale of the jet energy measurement, from comparisons of the mass peak position to 91 GeV; it would enable studies of jet energy resolution, by studying the width of the peak. All hot stuff! I was totally sold.
I started investigating what had been done by others in CDF. I soon found out that nobody was even dreaming of extracting a Z boson signal from dijet decays! I recall attending an informal meeting chaired by Weiming Yao – an esteemed colleague from Berkeley, who was a true expert in b-quark tagging (his was the standard algorithm used in CDF for that purpose) and later was to spend most of his time in the search for the Higgs decay to b-quarks. At the meeting I questioned him on whether there were plans to search for Z boson decaying into b-quark jets, and his reply was “it cannot be done. The background is too high!”. This motivated me even more to show CDF that it could be done. And if anybody could do it, I was sure that was me!: I was fully charged and confident, since I had just bagged a success in the reconstruction of another massive body from its fully hadronic decay: the top quark.
The first question to ask oneself in the search for decays was: where ? Not in the multijet sample! A two-jet dataset would be best, but dijet events are the most common result of a proton-antiproton collision, and they are the ones that the trigger usually discards rather than collect! Some trick had to be envisioned. What distinguishes b-jets from generic jets ? Their decay. B-quarks produce leptons, and electrons and muons are easily seen even in the few microseconds available for a trigger decision. That was it. I needed to study inclusive lepton datasets.
There followed three intense years. I blessed a first observation of decays in hadron collisions in June 1998 (see plot on the right: the inset on the upper right shows the dijet invariant mass distribution for the data (red points) with expected background overlaid in cyan, and the main graph shows the excess over background forming a nice gaussian peak sitting at about 90 GeV: the Z signal), and I could submit a contributed paper to ICHEP 1998, the international conference which was then held in Vancouver. But a real article on the search for Z boson decays to b-quark jets was not in the cards back then: I was too busy with other things in the next two years to allow myself a look back at Run I data: a new muon detector had to be built for Run II…
This has been the story of a failed paper. The next post will be the story of a paper which is finally published ten years afterwards, to get things right… Better late than never!